Researchers reported that they developed a new algorithm that more accurately predicts how long patients with advanced HF will survive, regardless of whether they receive a transplant.
“Our work suggests that more lives could be saved with the application of this new machine learning-based algorithm,” Mihaela van der Schaar, PhD, Chancellor’s Professor of Electrical and Computer Engineering at the UCLA Samueli School of Engineering, a fellow at the Alan Turing Institute in London, and the Man Professor at University of Oxford, said in a press release. “It would be especially useful for determining which patients need heart transplants most urgently and which patients are good candidates for bridge therapies such as implanted mechanical-assist devices.”
The algorithm, called Trees of Predictors, uses machine learning to account for 53 data points, enabling clinicians to compare patients who are awaiting transplant, and to optimize matching between donors and recipients.
The researchers analyzed 51,971 adult patients who received a transplant and 30,911 adults wait-listed for a transplant; all were included in the United Network for Organ Sharing database over 30 years.
The algorithm had an area under the curve of 0.66 for 3-month survival after transplant compared with 0.587 for the best clinical risk scoring method, the researchers wrote.
At 3 years after transplantation, the algorithm correctly predicted survival for 14% more patients vs. the best clinical risk scoring method when sensitivity was held at 80%, and correctly predicted mortality for 13% more patients vs. the best clinical risk scoring method when specificity was held at 80%, they wrote.
Similar improvements were seen for other time horizons and for pretransplantation predictions, according to the researchers.
“Following this method, we are able to identify a significant number of patients who are good transplant candidates but were not identified as such by traditional approaches,” Martin Cadeiras, MD, a cardiologist at the David Geffen School of Medicine at UCLA, said in the release. “This methodology better resembles the human thinking process by allowing multiple alternative solutions for the same problem but taking into consideration the variability of each individual.” – by Erik Swain
Disclosures: The authors report no relevant financial disclosures.